Of course, that's driven in part by advances in computation, just like a lot of other technology. Plasma physics, for example, which could yet make a big difference for energy.
Vernor Vinge predicted the singularity not before 2005 and not after 2030. Will we have crazysauce computers by then? Definitely. Will anything else in the world have caught up with the pace at which we apply transistors to wafers? Probably not.
One of Sterling's points is that the singularity is like dot-com startups in the 1990s: it doesn't have a business model. I think it would be right to say that it doesn't scale, either; there are too many other bottlenecks in the system where we wave our hands and say something about just throwing more metal at it.
Seriously, though, check out the talk, it's quite good.